This book provides an introduction to rule-based generative phonology for beginning undergraduates. The most immediately striking feature of the book is the sheer volume of realistic linguistic data it presents, from a wide range of languages, Bantu languages in particular. Another attraction is the large number of exercises. The datasets given in the exercises are extensive and minimally simplified, but there is a good range of difficulty. Very few of the exercises are trivial, but none should be beyond the capabilities of an intelligent undergraduate. A final feature of the book which makes it an excellent introductory text is the detail in which examples are worked through. Most phonologists have developed an intuitive understanding of how to extract generalisations from data, but this understanding can be difficult to articulate, for example, in a classroom setting. This book offers a way to address this, explicitly spelling out the train of thought required to arrive at a phonological analysis of each set of data.
Chapter 1 presents an overview of the scope of phonology as a field, distinguishing it from (acoustic) phonetics. The point is made that phonology is the study of sound structure as an abstract cognitive system rather than a physical phenomenon. An overview is given of speech segments and the properties which distinguish them, together with a brief summary in simple terms of the main topics of phonological study, introducing the study of phonological inventories (‘the sounds of a language’), phonotactics (‘rules for combining sounds’) and phonological alternations (‘variations in pronunciation’).
Chapter 2 introduces the notion of allophones and phonological rules which derive them. The concept of phonological contrast is introduced and illustrated using the (American) English stop system. Aspirated and unaspirated consonants are shown to be in a complementary distribution in English. The argument that the unaspirated form should be considered more basic in American English, and the aspirated form derived by rule, is explained. Other allophonic relations in the English stop system, such as those derived by flapping and t-glottaling, are also explored. Further examples of allophony are given from Tswana, Tohono O’odham, Kipsigis, Matuumbi, Kenyang, Sundanese, Mohawk, Ossetic and Makonde. The basic format of the phonological rewrite rule is introduced in the discussion of these examples.
Chapter 3 is an introduction to the breakdown of segments into binary distinctive features. The chapter begins with a discussion of possible contrasts across languages, arguing that a theory of phonology should allow us to recover constraints on possible contrasts. Similarly, the theory should be able to capture constraints on possible processes. It is explained that distinctive features are a way of formulating a theory which achieves these goals. An inventory of features is put forward, and the articulatory content of these features is discussed. The role of features in defining natural classes is explained, and it is shown how these natural classes constrain rules. The way in which a finite set of features (and implicational relationships within those features) limits the set of possible phonological inventories is also discussed. The chapter ends with a more formal account of the format of phonological rules and an explanation of how different feature theories have differing empirical implications, with a number of concrete examples, including problems with the acoustic feature [±flat], proposed by Jakobson & Halle (Reference Jakobson and Halle1956), the necessity of the feature [± labial] (Campbell Reference Campbell1974), and problems presented by apparent natural classes such as central vowels.
Chapter 4 focuses on the task of choosing the correct underlying representation from which surface forms are derived. The need for an abstract underlying form is motivated on the basis of neutralisation, illustrated using the example of final obstruent devoicing in Russian. Odden then gives a more precise definition of the underlying representation, as the output of the morphological component of grammar and the input to phonology. The remainder of the chapter largely focuses on methods to identify the correct underlying form – firstly on the basis of phonological alternations between different morphological forms of a lexical item, with examples given from Finnish, Kerewe and English (s-plurals). It is emphasised that underlying forms often deviate from citation forms, and the manner of deriving the correct underlying form from the observed distribution of segments is illustrated. Further datasets, from Chamorro, Korean, Koasati and Matuumbi, are given as practice exercises. Using additional examples from Korean and Matuumbi, it is demonstrated that sentence-level phonological alternations can also give insight into the identity of the underlying form of a lexeme. The chapter concludes by making the point that an underlying form cannot necessarily be deduced from a single morphological form of a word – exemplified with data from Samoan, Palauan, English (vowel reduction in unstressed syllables) and a very detailed worked example from Tonkawa.
Chapter 5 discusses interactions between phonological phenomena, overviewing how the different rule-orderings can recover different interactions. Feeding, bleeding, counterbleeding and counterfeeding interactions between rules are all illustrated (though these terms are not used). The separation of alternations into multiple rules is justified, and various examples are given of systems of rules, with simple feeding interactions, from Votic, Kamba, and Bukusu. Odden then goes on to illustrate the different possible interactions between rules, giving examples of bleeding in Lamba and Lithuanian, counterbleeding in New Julfa Armenian, and counterfeeding in Mongo. Further datasets for practice and discussion are given from Matuumbi, Karok, Shona and Klamath.
Chapter 6 could be considered the core of the book, providing a number of thoroughly worked analyses of complex data from a variety of languages. Odden runs through each step of forming a complete phonological analysis of a set of data, starting by developing a morphological breakdown, then collating phonological variants of each morpheme, before finally positing underlying forms and phonological rules. Odden emphasises that an individual phonological alternation may have multiple possible analyses, and that choosing between these analyses is possible only by considering the system of rules as a whole – metrics such as simplicity are only of use in choosing between analyses of a particular alternation if we reduce the complexity of the complete analysis. The examples Odden analyses are from Yawelmani, Hehe, Fore, Hebrew and Japanese.
Chapter 7 discusses the range of phonological phenomena which we find cross-linguistically, and which processes should be considered natural. There is a fairly brief discussion of the typology of phonological inventories, in which notions such as markedness and implicational relations are introduced. This is followed by a summary of processes which are common cross-linguistically, including various kinds of vowel harmony and consonant assimilation, lenition and fortition, segmental dissimilation, and other neutralising processes. These are each exemplified with data from a number of languages. Prosodically motivated processes, including epenthesis and deletion of vowels and consonants, lengthening and reduction conditioned by stress, and stress assignment are discussed. The chapter closes with a discussion of naturalness in phonology, reiterating the point that feature theory only allows the formulation of rules defined in terms of natural classes, and briefly mentioning similar constraints imposed by autosegmental analyses. Finally, Odden discusses rules which are formulable but largely unattested, focusing in particular on the coronalisation of labials. He explains that the difference in distribution between the very common process of velar coronalisation and the vanishingly rare process of labial coronalisation can be accounted for theory-externally: there is a simple, articulatorily grounded diachronic process which leads to the former but not to the latter.
Chapter 8 considers the degree of abstractness which is desirable in a phonological analysis, discussing a number of proposed cases of absolute neutralisation and motivating them in detail. The chapter opens by discussing why we might wish to rule out abstract analyses, with emphasis placed on the learnability issues involved if very abstract representations are assumed. It is then explained that a degree of abstractness does seem to be required to account for phonological alternations such as flapping – showing that abstract analyses should not be ruled out altogether, but should be kept to a minimum. Various proposals for restricting abstractness are discussed, such as a requirement that the underlying form of a lexeme must only contain segments which are present in some form of the lexeme, which amounts to a ban on absolute neutralisation, and which Odden rejects. Two examples which seem to require absolute neutralisation are discussed, from Matuumbi and Vedic Sanskrit. Odden then discusses even more highly abstract analyses, such as the analysis of the English vowel system given in Chomsky & Halle (Reference Chomsky and Halle1968): he offers arguments against this sort of analysis, pointing out in particular that the account is motivated on the basis of an unproductive, lexically restricted alternation. He then discusses how historical evidence can impact an abstract analysis, in particular discussing the case of rule-reversal in Yiddish final devoicing, which brings evidence to bear against abstract analyses of non-alternating segments. Further examples of absolute neutralisation are considered, from Yawelmani and Maltese – Odden points out that these are well-motivated, insofar as the proposed abstract segments simultaneously explain multiple phenomena. Finally, he discusses three cases where external evidence favours an abstract analysis, in Tera and Matuumbi (where the evidence comes from diachrony) and Bedouin Arabic (where the evidence comes from language games).
Finally, Chapter 9 provides a brief introduction to autosegmental phonology, focusing in particular on tonal phenomena. The autosegmental representation of contour tones is argued for, on the basis of the attested processes which these tones undergo. The notation for autosegmental rules is also introduced. Various tonal phenomena are discussed, including tonal stability and mobility, accros-the-board tone changes, inventories of complex tonal melodies, floating tones (and downstep), tonal morphemes and default tone assignment. All of these phenomena are used to argue in favour of an autosegmental model of tone. The extension of this model to non-tonal domains is discussed, illustrated using the example of pharyngeal harmony in Azerbaijani Aramaic. Finally, feature-geometric representations and non-linear models of syllabic structure are introduced and briefly exemplified. The chapter ends with a brief summary of non-linear elements not discussed in detail, including prosodic constituents other than the syllable.
There are a few minor issues with the book. At various points (particularly in Chapter 2) Odden discusses phenomena in English – the forms under discussion come from a variety of North American English, but this is not made clear in the text. For instance, when discussing the distribution of English aspiration (pp. 17–20), Odden asserts that English voiceless stops are unaspirated when they do not initiate a stressed syllable – a fact which does not hold of my variety of British English. Similarly, he discusses t-glottaling (pp. 21–22), which he takes only to appear before syllabic n. Again, he does not specify the dialect under discussion – the process applies much more freely in most British English varieties. Finally, the transcription of English vowels that Odden uses is distinctively American, consistently representing the LOT vowel as [a], which in my variety would be a better representation of the TRAP vowel. There is of course nothing to object to in the use of American English as a source of data, but without clarification that these properties are restricted to certain varieties, it seems likely that these points will produce confusion among speakers of non-American Englishes with no previous knowledge of phonology.
There are a few points at which typographical errors make the discussion difficult to follow, by rendering two opposing elements identically. For instance, on page 167 (example 35) the feature −nas is wrongly represented as +nas, on page 175 the sounds ʔ and ʕ are conflated (as ʔ), on page 207 ʃ and s are rendered identically (as ʃ), while c and ʧ both appear as ʧ on page 234. On page 287 (example 4, line 2) the features +H and −H are exchanged. Although the volume of data explored in this book is extremely impressive, there are occasional issues of detail. Tibetan, for instance, is transcribed in two different ways, with plain stops consistently transcribed as voiceless in one system (pp. 37, 219) and voiced in another (p. 112). This difference in transcription is not noted, which might lead to confusion. It is stated that voicing is non-contrastive in the language (p. 37), but this is true only because voicing is conditioned by tone, which is not transcribed here. Finally, Odden implies (p. 232) that labial coronalisation in front environments is not attested in any language – in fact, Tibetan (at least historically) exhibits just such a process. Hence Written Tibetan bya ‘(large) bird’ is pronounced [tʃʰà] in Standard Spoken Tibetan. The section on phonological typology comes rather late in the book – this is not a problem insofar as understanding the content of the chapters is concerned, but may present challenges when it comes to solving the problem sets. Unless a student has a good intuitive sense of the sorts of phenomena found cross-linguistically, they may well find it difficult to find a starting point to tackle the exercises, particularly those in Chapter 6, which involve rather complex interactions of multiple processes. Placing the discussion of phonological typology earlier in the book might have helped ameliorate this effect. Finally, the book gives no impression of the broader theoretical landscape. It is not really made clear that the approach to phonology offered is only one among many. No real mention is made of popular alternatives such as the various flavours of Optimality Theory, Government Phonology, Usage-based Phonology, etc. This runs the risk of giving the student a false impression that theoretical phonology is synonymous with rule-based generative phonology. It also means that the theoretical import of certain facts presented here is not really discussed – in Chapters 5 and 6 in particular, various complex, opaquely interacting processes are illustrated. The existence of opacity in phonology is of course evidence in favour of rule-based phonology against alternatives such as (Classic) OT, but the word opacity is never even mentioned, let alone the fact that it is the topic of much current theoretical work. In the end, however, these are all relatively trivial points, particularly in an elementary textbook of this type: this volume remains, in my view, the best introductory text on phonology available.
